Professional Documents
Culture Documents
A Fundamental Study of The Flow Past A Circular Cylinder Using Abaqus/CFD
A Fundamental Study of The Flow Past A Circular Cylinder Using Abaqus/CFD
1. Introduction
We have studied the fluid dynamics in basic engineering education. However, it is not particularly
easy to apply knowledge of fluid dynamics in to actual designs. In practice, although structural
mechanics may appear in every aspect of the design process, it is very rare for fluid dynamics to
be present in the apparent in the original equations. In many cases, fluid flow is translated using
particular design formulae. Pressure drop calculations in a pipe line or the calculation of the heat
transfer coefficient are examples of typical formula-based design.
The environment contains a multitude of various fluids. Accordingly, there are plenty of objects to
which fluid dynamics can be applied. However, not only are fluid flows not visible, it is very
difficult to understand the following behaviors of fluids:
i. Fluid is significantly deformable compared with solids, and the degree of deformability
depends upon the viscosity of the fluid.
ii. The interaction of a fluid flow with a solid surface is important subjects. More
specifically, the attachment to a solid surface, the development of a boundary layer, and
the separation.
iii. In addition to these phenomena, the transition of laminar flow to turbulent flow occurs.
These properties of fluid flow make it difficult to apply fluid dynamics in practice. Abaqus is
considered to be the best option for general purpose nonlinear FEM, and the latest version
includes Abaqus/CFD. Abaqus/CFD is a brand new program developed within the Abaqus system,
Again, considering the role of fluid dynamics, the role can be categorized into the following three
functions:
(a) to understand actual fluid motion ;
(b) to identify the fluid motion induced force acting on an object immersed in a fluid;
(c) to determine how an object moving in a fluid and subjected to fluid-induced force
changes its motion and how such a moving object, in turn, simultaneously causes a
change in the motion of fluid.
In terms of utilizing CFD in turbulence modeling or in development new analytical approach ,
such as the particle method, (a) is very important. In addition to comprehensively deal with the
fluid-structure coupling problem, it is important to address (c). However, in relation to common
aspects of mechanical design, (b) is the best target for exploiting CFD.
In general, textbook on fluid mechanics provide examples fluid flow around a circular cylinder as
basic problems. Taking such a problem, the fluid-induced force acting on a circular cylinder
immersed in a uniform flow is estimated. However, the problem is addressed not in a single
chapter, but over several chapters. For example, a chapter on ideal fluids may describe the
D’Alembert’s paradox and periodical characteristics of the von Karman vortices, and another
chapter on viscous flows then describes the separation of boundary layers. The relationship
between momentum and drag is elaborated upon a chapter of its association with the formation of
wake.
As a result, it is almost impossible to develop a comprehensive understanding of the above-
mentioned behaviors and characteristics of fluid flow based on the knowledge in university
textbooks. If CFD is used following proper guidance in appropriate tutorials, it will be possible for
a mechanical engineer to analyze the problem of fluid flow around a circular cylinder with relative
ease. Implementing CFD into structural engineering software may open a new way toward
understanding complicated fluid flow behaviors, something that would substantially benefit many
engineers.
We performed a series of analyses in which fluid flow around a circular cylinder immersed in a
uniform flow was used to trace the occurrence of von Karman vortices and to determine the
resultant vibration of the cylinder.
Figure 1-(a) shows a flow with a very small Reynolds number. The shape of every streamline is
symmetrical not only around the cylinder’s upper to lower side, but also around its front to rear.
As the Reynolds number increases, the front to rear symmetry disappears, and the interval of the
streamlines of the cylinder’s rear flow widens. As the Reynolds number reaches close to 10, a
closed region of the streamline is generated within the cylinder’s rear flow. The fluid in the upper
half of this region rotates clockwise, and the fluid in the lower half rotates counter clockwise. This
phenomenon is the result of the so-called generation of vortices. Figure 1-(b) depicts this state.
As the Reynolds number increases, the length of the vortex increases. When the Reynolds number
exceeds 60, the rear flow becomes unstable as shown in Figure 1-(c), and the vortex begins to
oscillate in an up-down direction. The fully formed vortex is carried away on the surrounding
main flow, and a vortex subsequently occurs on the other side. This vortex then expands before
finally dissipating and departing leaving the circular cylinder. In addition to this flow process,
every pair of upper and lower vortices behind the cylinder is stored in an alternative regular array
to form twin rows of vortices. These rows are called the von Karman vortex street. While the main
flow approaching the circular cylinder is in a uniformly steady state, a periodically varying flow
characteristic of the von Karman vortex street appears in the cylinder’s rear flow.
As shown in Figure 1-(d), as the Reynolds number reaches 1,000, the vortices are mixed together,
and the flow behind the cylinder (wake) behaves very irregularly in time and space scale. This
behavior is the so-called state of turbulent flow. However, a free stream distant from the wake
region maintains a steady state, and the streamlines are held in a regular and smooth shape. This
phenomenon is known as laminar flow.
Abaqus/CFD was used in this study to perform analyses, within a range of Reynolds numbers up
to 200 under which we considered it possible to analyze the occurrence of laminar flows.
4. Analysis results
4.1 Variation in flow patterns corresponding to various Reynolds number
Using a dual-core CPU, the analysis of CASE1 took about 60 sec. The time taken to perform the
analysis duration increased with an increase in the Reynolds number, with the analysis of CASE8
taking about 8,000 sec.
Streamlines derived from the analysis are shown in Figure 3. Some typical cases compared with
the experimental results (Van Dyke, 1982; JSME, 1992) are shown in Figure 4.
CASE1 and CASE2 correspond to flows with a sufficiently low Reynolds number (slow flow),
and their streamlines appear symmetrically, front to rear. When the Reynolds number exceeded 1,
as seen in CASE3, the symmetry was lost in the front and rear flows. However, no separation of
fluid flow occurred. CASE4 to CASE6 correspond to a Reynolds number in the range of 20 to 50,
and a pair of upper and lower vortices was generated within the wake of the cylinder. As the
Reynolds number increased, the vortex region expanded and increased in length.
(e) CASE5 Re = 26
(f) CASE6 Re = 50
Figure 6 shows the distributions of the velocities (velocity measured at 0.1 mm from the surface),
pressures and wall shear stress at θ = 0 ~ 180 deg on the cylinder surface. As the stagnation point
is situated at the front edge of the cylinder, the velocity was zero. The velocity of the cylinder’s
front flow accelerated, whereas the velocity of the rear flow decreased. According to Bernoulli’s
principle, there is a certain relationship between flow velocity and pressure, with increasing flow
velocity resulting decreasing pressure. The resultant pressure distributions conformed to this
principle.
As the flows moved down-stream and their respective velocities were gradually reduced due to the
effects of viscosity and the pressure gradient, the flow can no longer travel along the surface of the
cylinder. This phenomenon is called the separation point, which is the point at which the flow is
separated from the cylinder’s surface. In Figure 6, the point location θ ≈ 130 deg corresponds to
the separation point at which the wall’s shear stress became zero.
0.8 2.5E-08
Velocity
0.7 2.0E-08
0.6 Pressure 1.5E-08
Velocity (mm/sec)
0.3 0.0E+00
0.2 -5.0E-09
0.1 -1.0E-08
0 -1.5E-08
-0.1 -2.0E-08
0 20 40 60 80 100 120 140 160 180
Angle (deg) 130 (deg)
Figure 6. Velocity on the cylinder’s surface, distribution of pressure,
and the wall’s shear stress
4.0E-10 6.0E-06
3.5E-10 CASE8
CASE1 5.0E-06
3.0E-10 (Re=195)
(Re=0.038)
Fluid force (N)
4.0E-06
2.5E-10
2.0E-10 3.0E-06
1.5E-10
2.0E-06
1.0E-10
1.0E-06
5.0E-11
0.0E+00 0.0E+00
0 50 100 150 200 250 300 0 50 100 150 200
Time (sec) Time (sec)
(a) CASE1 Re = 0.038 (b) CASE8 Re = 195
Figure 7. Time history curves of fluid force (CASE1, CASE8)
The drag coefficients in each analysis calculated using Equation 3, the drag coefficients obtained
from the experiments (Hughes and Brighton, 1999), and the results obtained using the Oseen
approximation (Equation 4) by Lamb are shown together in Figure 8.
8π
CD = (4)
Re(2.002 − ln Re)
103
1000
Experiment[1]
102
100 Oseen's
approximation
Abaqus/CFD
Analysis result
Cd ( - )
10
10
11
0.1-1
10
10-1 1.0E+00
1.0E-01 1 102 1.0E+03
10 1.0E+02
1.0E+01 103 1.0E+04
104 1.0E+05
105 1.0E+06
106
Re ( - )
⎛ 21.1 ⎞
S t = 0.212 × ⎜1 − ⎟ (7)
⎝ Re ⎠
0.16 Experiment 5
Experiment 6
0.14
Rayleigh
0.12 Roshiko
Analysis resu
Abaqus/CFD
0.1
20 40 60 80 100 120 140 160 180 200
Re ( - )
The Strouhal numbers Stn vs. the vibration frequencies of the cylinder are plotted in Figure 10.
Figure 11 shows the vibration amplitudes Y of the cylinder vs. the various Strouhal numbers Stn. In
both of these figures experimental results (Hartlen, 1970) and our analytical results are plotted.
As clearly noted in Figure 10, a lock-in phenomenon is observed in the experimental results within
a range of Stn = 0.8~1.15, which indicates the existence of a specific region where vibration
frequencies of the cylinder remain mostly in a similar range. However, it was not possible to
visualize this lock-in phenomenon in this analysis. Therefore, the results show a tendency of
monotonous increase in the vibration frequencies of the cylinder in accordance with an increase in
the Strouhal number. If a cylinder is excited by von Karman vortices, it is likely that locally three-
dimensional flow in the vicinity of the cylinder might occur. However, the model in our analysis
was not capable of tracing such phenomena.
In addition, as seen in Figure 11, the vibration amplitudes of the cylinder sharply increased in the
lock-in region, something that would lead to a high possibility of the flow transitioning into
turbulent flow. This study was largely limited to laminar flow; the possibility of turbulent flow
should be investigated in future studies.
6. Conclusion
Taking fluid flows generated around a circular cylinder immersed in a uniform flow as an object
of study, a series of analyses using Abaqus/CFD were performed for a specific region with
Reynolds numbers up to 200. The results were found to be in good agreement with those of earlier
experiments. As a consequence, we have obtained comprehensive knowledge and findings that can
be applied to problems associated with fundamental estimation of the flow around the circular
cylinder. Although we verified the basic capabilities of Abaqus/CFD with regard to coupled
structural fluid analysis in this study, additional analyses are needed to verify more advanced
features of the software.
0.6
Dimensionless Amplitude [-]
0.5
Experiment
0.4
Abaqus/CFD
0.3
0.2
0.1
0
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
Strouhal Number [-]
7. References
1. Hughes, W.F., & Brighton, J.A., “Schaum’s outline of fluid dynamics,” USA: McGraw-Hill,
p.102, 1999.
2. Van Dyke, M., “An album of fluid motion,” USA: The Parabolid Press, p.28, 1982
3. The Japan Society of Mechanical Engineers. “Photograph collection of Flow,” JPN: Maruzen
Company, pp.2-5, 1992.
4. Zdravkovich, M.M., “Flow around circular cylinders, Vol.1: Fundamentals,” USA: Oxford
University Press USA, p.61, pp.258-259, 1997.
5. Hartlen, R.T., & Currie, I.G., “Lift-oscillator model of vortex-induced vibration,” Proc. ASCE,
J. Eng. Mech. Div, Vol.96, No.5, pp.577-591, 1970